Affiliations: [a]
Department of Electrical Engineering, Faculty of Engineering and Technology, Jamia Millia Islamia, Delhi, India
| [b]
Defence Institute of Physiology and Allied Sciences, DRDO, Delhi, India
Correspondence:
[*]
Corresponding author. Md Qaiser Reza. Department of Electrical Engineering, Faculty of Engineering and
Technology, Jamia Millia Islamia, New Delhi, India. E-mail: rezaalam2004@gmail.com.
Abstract: The acoustic resonance spectroscopy is an accurate, precise, inexpensive, and non-destructive method for identification and quantification of materials. The acoustics based inspection methods used for classification of materials in the field of food, security, and healthcare is constrained by expensive instrumentation, complicated transducer coupling, etc. Hence, a simple, inexpensive, and portable system has been devised that acquires data quickly and classifies the materials. It has two piezoelectric transducers glued to both ends of the V-shaped quartz tube, one acting as a transmitter and another as a receiver. The transmitter generates vibration by white noise excitation. The receiver detects the resultant signal after interaction with samples and recorded the acoustic signal with the help of a laptop and software. From analysis of power spectrum of signals acquired from each of the samples, seven resonant peaks were obtained. PCA analysis was carried out by selecting only two principal components as feature vectors for classification. The overall accuracy of the classifiers: LDA and Naive Bayes were 98.91% and 96.83% respectively. The classification accuracy of LDA for distilled water, sugar solution, and salt solution were found to be 100%, 98.5%, and 98.25% respectively, while the accuracy of the Naive Bayes classifier was 94%, 98.5%, and 98% respectively. The results show that the classification accuracy of LDA is better than Naive Bayes classifier. The datasets of the developed simple system show a significant capability in the classification of materials.
Keywords: Acoustic resonance spectroscopy (ARS), acoustic signature, principal
component analysis (PCA), linear discriminant analysis (LDA)
